ADMISSIBLE HEURISTIC FUNCTIONS FOR FLOW-SHOP PROBLEMS

摘要

A set of heuristic functions is developed for use in conjunction with an intelligent heuristic search algorithm (IHSA~*) for solving flow-shop problems involving two or three machines with any number of jobs. The admissibility of these functions is established and a subset of three admissible heuristic functions is identified in terms of the close proximity of their values to the minimum makespan of the problem. Experimental results indicate that, choosing among these three admissible heuristic functions, the one with the largest value improves performance characteristics of IHSA~*. In addition, for the particular case where any one of the machines dominates the others, it is shown that exactly one of this subset of three heuristic functions can be identified immediately as having a value that is significantly closer to the minimum makespan than the others, and experimental results again illustrate the resulting improvement in the performance of IHSA~*.